In the following paragraphs, we describe some examples of applications of pStore other than a digital movie studio to demonstrate the generality of our proposal.
Online data sharing. In general, it is desirable that each object can have an arbitrary metadata structure suitable for describing its contents as well as its relationships with other objects. Objects can relate to each other in many different ways: an object may overlap with or include other objects; multiple objects may share descriptive data. In practice, meaningful objects are often identified and associated with their descriptive data incrementally and dynamically, after the data is stored in the system.
To provide adequate control, users can be given different access privileges. To facilitate collaboration, in addition to a shared global view of all the data, there may also be customized per-user and per-application views. Advanced searching capabilities are needed to allow people to effectively navigate among the various digital components.
A semantic, deep archival system.
It is now practically affordable to archive each individual version of a file. Such archival storage system are becoming essential for many critical applications. We list some desirable features.
First, a user would like the file store to have a ``travel-in-time" capability---every change to an object or to the name space is recorded, and a user can travel arbitrary back in time to retrieve any version of a file that ever existed [11]. An important challenge is to maintain the various dependencies among different versions of objects and handle time as yet another type of semantic information.
Second, to reduce storage space consumption, objects should be stored efficiently. Various data clustering and compression techniques are being explored. One way to do this is to exploit the available semantic information. E.g., when generating a new version of a file, the semantic information is used to identify an existing (base) file with similar contents. Only the differences between the new and the base file are stored.
Last, in restoring a backed-up version, the biggest headache is to find the right document and the right version. With pStore 's rich metadata model, the semantic information of files can be associated with files. In the restoring operation, the user describes a desired feature that is known to exist in the recovered version. For example, the system may use content extracts to locate the right version, without requiring the user remembering the exact name or creation date of the restored file.
Digital content distribution. In addition to search capabilities, a large-scale distributed file system can utilize the relationships among files to guide data placement, and perform caching and prefetching. CDN more efficient. Another related application is to support data hoarding for mobile users. Before disconnected from the network, all frequently used data for the user are identified through examining the metadata, and are automatically moved to a portable device. Systems such as SEER [10] use simple semantic hints such as user activity and directory membership for hoarding related files. Their effectiveness is limited by operations such as running the UNIX find utility across an entire file system.
Personal storage for desktop users. Many of the features described above can benefit ordinary desktop users as well. As desktop users, we would like to keep every version of important files that we ever created or downloaded, add arbitrary annotations to the files, relate them to the their sources, and create cross links among them. Automated file hoarding can relieve much of the pain to manually identify and move files among computers and mobile devices. Many of us have painful experiences of not finding files. The advanced searching capability would make search much easier.